Correct stale architecture docs in experimental/ssh/README.md - #6451
Open
anton-107 wants to merge 1 commit into
Open
Correct stale architecture docs in experimental/ssh/README.md#6451anton-107 wants to merge 1 commit into
anton-107 wants to merge 1 commit into
Conversation
The compute requirements claimed terminal SSH works on standard access mode. It does not: ValidateClusterAccess requires Dedicated access mode with single_user_name set, and both entry points call it (client.go on a direct `connect --cluster`, setup.go on `ssh setup`), so a standard, dedicated-to-a-group, or no-isolation cluster is rejected before any work happens. The two design diagrams had drifted from the code as well: - Keys travel through a per-session secret scope, not a pub-key baked into the bootstrap notebook. The notebook is `ssh-server-bootstrap`, not `start-server-with-pub-key.ipynb`. - Workspace paths are /Workspace/Users/$me/.databricks/ssh-tunnel/$v/..., not ~/.ssh/$v/..., and metadata.json also carries cluster_id and usage_policy_id. - The server serves /logs alongside /ssh and /metadata, publishes metadata before it starts accepting connections, and never deletes metadata.json - it just exits, which is why the client must re-probe /metadata through the driver proxy rather than trusting the file. - The ProxyCommand takes --cluster/--metadata flags, and the websocket URL includes /o/$workspaceId and ?id=$connId. - The high-level diagram showed per-Spark-user servers, an arrangement that only made sense while standard access mode was in scope. Co-authored-by: Isaac <no-reply@databricks.com>
rclarey
approved these changes
Sep 2, 2026
rugpanov
approved these changes
Sep 2, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Changes
Docs only, one file. Corrects
experimental/ssh/README.mdwhere it had drifted from the code:ValidateClusterAccessrequires Dedicated access mode withsingle_user_nameset, and both entry points call it, so standard / dedicated-to-a-group / no-isolation clusters are rejected before any work happens. Also mentions serverless, which the list omitted entirely./logsendpoint, theProxyCommandflags, and the websocket URL shape.DELETE metadata.jsonon teardown. The server just callsos.Exit(0)and leaves the file behind.metadata.jsonis published before the listener accepts and outlives the process, so neither its presence nor its contents prove a server is running. That is why the client always re-probes/metadatathrough the driver proxy.The high-level diagram previously showed per-Spark-user SSH servers, an arrangement that only made sense while standard access mode was in scope.
Why
Someone reading this README to learn the feature would come away with the wrong mental model in three ways that matter: they would expect standard-mode clusters to work, look for
metadata.jsonunder~/.ssh/, and assume a stale metadata file means no server is running.Verified each claim against
origin/mainrather than trusting the prose:ValidateClusterAccessandsetup.go,internal/workspace/workspace.gofor paths,internal/server/server.gofor endpoints and the exit path,internal/keys/for key flow, andbuildSSHArgs/ToProxyCommandfor the spawned command.Tests
No tests - docs-only, no behavior change.
./tools/validate_whitespace.pypasses andgo mod tidyleaves the modules untouched. No changelog fragment:experimental/and doc-only both fall under the skip list in the pr-checklist skill.I could not render the Mermaid locally (no npm registry access from this host), so I validated the diagrams structurally instead - subgraph/
endbalance, no undeclared node or participant references, every arrow carries a label - and confirmed the new text introduces no character that the previously-rendering diagrams did not already use. Worth eyeballing both diagrams in the GitHub preview before merging.This PR was written by Claude Code.